ERC Starting Grant: 

Daniel Garrett

Dynamic Mechanisms (DYNMECH).  Grant agreement No. 714147.  January 2017 to June 2023.

This project is funded by the European Research Council and has most recently been hosted by the University of Essex.  The project has ended (ended on 30 June, 2023).

This project aims at contributions to the literature on dynamic mechanisms.  A dynamic mechanism is a long-term agreement between a principal and agents determining allocations, actions, and payments over time, often as a function of the agents’ choices. Applications include the selling or trading of goods, long-term employment relationships, regulation and public finance. That the environment is “dynamic” typically means that it changes with time: for instance, new players arrive or the preferences of existing players evolve, often randomly.

Four principal areas are considered.

Competition.  Competition in contracts is often regarded as difficult to model because agreements can potentially condition on rich information such as the contracts offered by other players.  By restricting the contracts that players can offer, one obtains predictions that seem more informative about many real-world markets.   An example is the classic dynamic setting of Burdett and Mortensen (1998) where agents (workers) search for job offers over time.  In [3], we have extended the setting to cases where agents have private information on their preferences (for instance, this could be preferences for working hard which would often be privately known by the worker).  We provided further insights on player welfare for this problem in [6].  In [13], I study sellers who can choose their future offers flexibly and who can fully commit to them, while the agents (here, buyers) encounter the competing sellers over time.  Equilibrium turns out to coincide with the one when sellers cannot commit to future offers, but do follow "best price policies" which are guarantees to refund buyers who find better offers.  This provides one justification for studying commitment, as best price policies are promises made by many retailers in practice.  In all these works, competition leads to randomization over offers or prices, which has been connected in the literature to price dispersion and also "sales".  In [10], we propose a dynamic theory of random sales based on buyer risk aversion which provides an alternative to the predominant theory of random prices based on competition and search frictions. 

Robust predictions in dynamic mechanism design.  An important part of the literature on dynamic mechanism design considers optimal agreements when agents' private information evolves randomly with time.  Most of the existing literature is restricted to a narrow class of stochastic processes that proves sufficiently tractable.  In [11], we develop alternative approaches for characterizing optimal dynamic contracts for a much broader class of stochastic processes.  This is possible because we ask only for a partial characterization of optimal contracts.  In [5], I propose an approach to making partial predictions on welfare in static incentive problems where agents may have different preferences for exerting effort unknown to an outside observer.  The work anticipates application to a class of problems in dynamic mechanism design, which is the subject of ongoing work ([14]).

Bilateral trade with budget balance and dynamic arrivals.  The classic ``bilateral trade'' problem with budget balance (Myerson and Satterthwaite, 1983) shows that the initial allocation of property rights is generally crucial even when trade is allowed. Repeated interaction was subsequently shown to provide a way around Myerson and Satterthwaite's impossibility result.    In [7], I show that the impossibility result is restored, essentially in its original form, if (realistically) agents arrive to the market over time and have private information on their arrival date.  

Investments by agents under dynamic contracts.  An important consideration for dynamic contracting is the investments that can be made by agents over time:  either as stipulated by the contract or outside the relationship.  In  [8], we study so-called "relational contracts" when agents (workers) can invest in outside options (by saving) either publicly or privately.  In [12], we study dynamic screening when the agent is hired to make investments and has private information on both cost and usefulness to the procurer.  In [9], we study investments by the agent in the production technology given that the agent will be subsequently contracted under moral hazard.  These works can help understand a range of applications that are inherently dynamic where there are both investments and contracting.


My papers supported by the grant

Published papers

Working papers

Work in progress